Prepare your search implementation

AI Personalization is a beta feature according to Algolia’s Terms of Service (“Beta Services”).

Once you’ve prepared your event implementation, focus on your search implementation.

Attach user tokens to queries

To ensure that AI Personalization can understand each user’s journey and continuously improve their personalization experience, attach a user token to each query. The user tokens attached to the queries must match the user tokens sent with events.

Attach user tokens with InstantSearch

To include the userToken in your search queries using InstantSearch, use the configure widget to set the userToken search parameter.

1
2
3
instantsearch.widgets.configure({
  userToken: 'user-1234',
});

1
searcher.request.query.userToken = 'user-1234'

1
2
3
<ais-configure
  [searchParameters]="{ userToken: 'user-1234' }"
></ais-configure>

1
<ais-configure userToken="user-1234" />

1
<Configure userToken="user-1234" />

1
searcher.query.userToken = 'user-1234'

1
2
3
4
5
final hitsSearcher = HitsSearcher.create(
  applicationID: 'YourApplicationID',
  apiKey: 'YourSearchOnlyApiKey',
  state: const SearchState(userToken: 'user-1234'),
);

Attach user tokens with an API client

To include the userToken in your search queries with an API client, use the search method to send the userToken search parameter.

1
2
3
$index->search('query', [
  'userToken' => 'user-1234'
]);

1
index.search('query', { userToken: 'user-1234' })

1
2
3
4
5
index.search('query', {
  userToken: 'user-1234'
}).then(({ hits }) => {
  console.log(hits);
});

1
index.search('query', {'userToken': 'user-1234'});

1
2
3
4
5
6
7
8
let query = Query("query")
  .set(\.userToken, to: "user-1234")

index.search(query: query) { result in
  if case .success(let response) = result {
    print("Response: \(response)")
  }
}

1
2
Query query = new Query("query");
query.UserToken = "user-1234";

1
2
Query query = new Query("query")
.setUserToken("user-1234");

1
2
3
4
res, err := index.Search(
  "query",
  opt.UserToken("user-1234"),
)

1
2
3
4
5
6
client.execute {
  search into "myIndex" query Query(
    query = Some("query"),
    userToken = Some("user-1234")
  )
}

Persist user tokens

For user profiles to be found and retrieved at search time, you should persist anonymous user tokens and authenticated user tokens across sessions.

Avoid enabling personalization at query time

Rather than using the enablePersonalization search parameter at query time, set it in the index settings when moving to production. This lets you A/B test the impact of personalization without having to modify code.

Next step

Now that you’ve prepared your index structure, event implementation and search implementation, AI Personalization can build a holistic view of each user’s journey and personalize search results accordingly.

You’re ready to configure AI Personalization indices without additional code changes.